A Specific Domain of G i (cid:1) Required for the Transactivation of G i (cid:1) by Tubulin Is Implicated in the Organization of Cellular Microtubules*

G s (cid:1) , G i (cid:1) 1 , and G q (cid:1) subunits bind tubulin with high affinity, whereas transducin (G t (cid:1) ) does not. The inter- action between tubulin and G (cid:1) , which also involves the direct transfer of GTP from tubulin to G (cid:1) (transactiva-tion), is not yet fully understood. This study, using chimeras of G i (cid:1) and G t (cid:1) , showed that the G i (cid:1) (215–295) segment converted G t (cid:1) to bind to tubulin and this chi - mera (chimera 1) could be transactivated by tubulin. Insertion of G t (cid:1) (237–270) into chimera 1 to form chi- mera 2 resulted in a protein that, like G t (cid:1) , did not bind tubulin. Thus, it was thought that the G i (cid:1) (237–270) do- main was essential to modulate the binding of G i (cid:1) 1 to tubulin. Surprisingly, when domain (237–270) of G i (cid:1) was replaced by G t (cid:1) (237–270) to form chimera

G proteins act as intracellular transducers to propagate a variety of signals across the plasma membrane. Due to their interaction with transmembrane receptors and lipid modification, G proteins are usually associated with the plasma membrane. Recently, increasing evidence has emerged that G proteins are also present at intracellular areas such as Golgi apparatus, endoplasmic reticulum, cytoskeleton, and even the nucleus (1)(2)(3)(4)(5). Moreover, some studies have also shown that activated G␣s can be released from the plasma membrane to the cytoplasm (6 -9). Thus, it is possible that G proteins exist in several different cellular locations and play roles in various physiologic processes.
Microtubules, a major component of the cytoskeleton, are involved in many cellular functions including chromosome movements during mitosis, intracellular transport, and the modulations of cell morphology. The biological function of microtubules is based, in significant part, on the ability of tubulin to polymerize and depolymerize. A heterodimer of ␣and ␤-tubulin is the basic building block of microtubules. Tubulin is a GTP-binding protein, two GTP molecules are bound noncovalently in exchangeable (E-site on ␤-tubulin) and nonexchangeable (N-site on ␣-tubulin) sites. Both G␣ subunits and tubulin have intrinsic GTPase activity but that of tubulin is activated during the process of polymerization (10) or when complexed with G␣ (11).
Certain G protein ␣ subunits (G s ␣, G i ␣ 1 , 1 and G q ␣) bind to tubulin with high affinity (K D Х120 nM). Transducin, however, does not bind to tubulin. Complexes between G s ␣ or G i ␣ and tubulin were co-immunoprecipitated from detergent extracts of synaptic membrane (12)(13)(14)(15). This binding appears to activate the G protein due to a direct transfer of GTP from the E-site in tubulin to G␣ (transactivation) (16 -18). In addition, the association of G␣ and tubulin induces GTPase activity in tubulin and modulates microtubule polymerization dynamics in vitro (11,19). Agonists for certain G protein-coupled receptors induce microtubule depolymerization and tubulin relocation to the plasma membrane (20). These studies suggest that microtubules might be involved in regulation of signal transduction, whereas G protein signaling may modulate microtubule polymerization.
In the present study, functional domains in the G i ␣ 1 amino acid sequence that mediate the interaction of G i ␣ with tubulin were investigated. We report here that the 237-270 region of G i ␣ appears not only to contribute to the binding of G i ␣ to tubulin but also is crucial for the transactivation of G␣ by tubulin. The transactivation of G␣ by tubulin may be an important factor for the modulation of tubulin polymerization dynamics and cell shape by hormone and neurotransmitters. mera B, (G t ␣ 1-236/G i ␣ 1 236 -350) were first constructed. These were used to construct chimera 3 by replacing the chimera A C-terminal sequence (270 -350) with the chimera B G i ␣ 1 (270 -350) domain. To assemble DNA fragments encoding chimera 3, both plasmids were digested with HindIII and AsuII. A short AsuII-HindIII DNA fragment from the expression vector encoding chimera B was ligated with a large DNA fragment derived from the expression vector encoding chimera A. This new chimeric G t ␣/G i ␣ 1 construct, chimera 3, was confirmed by restriction analysis and DNA sequencing. Chimera 1 and chimera 2 had been previously described (21). In addition, all chimeras and His 6 tag G i ␣ 1 were subcloned into PcDNA3 vector (Invitrogen) for expression in the mammalian cells.

Construction of Chimeras
Expression and Purification of Chimera 3 and G i ␣ 1 -The Escherichia coli BL21 cells transformed with the vectors harboring chimera 3 and G i ␣ 1 were grown in 2ϫ YT medium with 100 g/ml of ampicillin at room temperature up to A 600 of 0.5 and then induced with 30 M isopropyl-1-thio-␤-D-galactopyranoside at room temperature for 16 -20 h. The cell pellet was resuspended in 1:20 of a cell culture volume of buffer containing 50 mM Tris-HCl, pH 8.0, 50 mM NaCl, 5 mM MgCl 2 , 50 M GDP, 0.1 mM phenylmethylsulfonyl fluoride, and 5 mM ␤-mercaptoethanol (buffer A) and disrupted by ultra-sonication using eight 30 s pulses with 1 min breaks in between pulses. The crude cell lysate was cleared by centrifugation at 100,000 ϫ g for 60 min. The supernatant was collected and adjusted to 20 mM Tris-HCl, 500 mM NaCl, and 20 mM imidazole (binding buffer). The cell lysate was loaded onto 5 ml of nickel-nitrilotriacetic acid column preliminary equilibrated with the binding buffer. After washing the column with 10 bed volumes of binding buffer, the bound material was eluted with 20 mM Tris-HCl, 500 mM NaCl, and 100 mM imidazole and subjected to overnight dialysis against buffer A with 20% glycerol. The protein samples were directly applied to 1 ml Mono Q column equilibrated with buffer A without GDP and ␤-mercaptoethanol. Proteins were eluted with 0 -1 M NaCl gradients in buffer A. An increase in tryptophan fluorescence was measured with excitation at 280 nm and emission at 340 nm to monitor AlF 4 Ϫ -dependent conformational change of chimeric G␣ subunits, to determine the activity of the purified protein as described by Skiba et al. (21). The eluted proteins were supplemented with GDP, ␤-mercaptoethanol, and phenylmethylsulfonyl fluoride to a concentration of 25 M, 2 mM, and 0.1 mM, respectively, aliquoted and stored at Ϫ80°C for several months with no loss of functional activity (21).
AlF 4 Ϫ -dependent Conformational Change of G␣-Binding of AlF 4 Ϫ to G␣-GDP mimics the GTP conformation of the molecule and changes in Trp fluorescence can be used to monitor the "viability" of a G␣ construct. To measure this, 200 nM G i ␣ 1 or the chimeras were incubated at room temperature in 50 mM Tris-HCl, pH 8.0, 50 mM NaCl, and 2 mM MgCl. Fluorescence was measured in an Aminco-Bowman Series 2 Spectrometer (SLM-Aminco) using excitation of 280 nm and emission at 340 nm. Measurements were taken before and 5 min after the addition of 10 mM NaF and 30 M AlCl 3 . Fluorescence increases were expressed as a percent change from the initial fluorescence: (⌬F (%) ϭ (FϪF o )/F o ϫ 100, where F o is the initial fluorescence and F is the fluorescence after fluoride addition. This method is described in Ref. 21.
Tubulin Iodination-Ovine brain tubulin was made by two assembly-disassembly cycles (22), in the presence of microtubule-associated proteins, which were subsequently removed by phosphocellulose chromatography. An aliquot of 100 g of PC-tubulin in 100 l of PIPES buffer was applied to a 12 ϫ 75 mm glass tube precoated with 100 g IODO-GEN (Pierce), which was dissolved in 100 l of triethanolamine and dried in a ventilated hood. 2 Ci of Na 125 I (Amersham Biosciences) was then added, and the reaction was allowed to proceed with gentle agitation for 15 min. The reaction was terminated by addition of 100 l of PIPES buffer containing 4 mM dithiothreitol. The free iodide was removed by ultrafiltration by loading the tubulin suspension on a 3-ml P-6DG (Bio-Rad) desalting column twice. The desalted 125 I-tubulin was centrifuged at 11,500 rpm for 10 min to remove denatured protein, and the supernatant containing iodinated tubulin was used in protein binding experiments (14).
Binding of Tubulin to G␣-Purified G i ␣ 1 , chimera 3, bovine transducin, and ovalbumin (Sigma) were applied to nitrocellulose membrane (Midwest Scientific) in the amounts of 150, 100, 50, 25 ng, respectively. Nitrocellulose was incubated with 10% bovine serum albumin in 100 mM PIPES buffer at room temperature for 2 h to block nonspecific binding and then was incubated with 100 nM iodinated tubulin in 100 mM PIPES buffer at room temperature for 2 h. Nitrocellulose was then washed three times with PIPES buffer. Binding of tubulin to G protein ␣ subunits was detected by radioautography. The affinity of tubulin for chimera 3 was estimated by immobilizing chimera 3 upon nitrocellulose and hybridizing with varying concentrations of 125 I-tubulin. This was quantified in a gamma counter (Beckman Gamma 9000). The binding was dose-dependent and saturable (14). Immobilized chimeras (other than chimera 2) bound 0.3-0.5 mol of tubulin/1 mol of G␣.
Transfection-Plasmids were purified using the Qiagen Maxi purification kit. COS-1 cells were split and plated in a 1:15 dilution to 10 cm plates the day before transfection. Cells were transfected either with calcium phosphate or Lipofectin (Invitrogen). For calcium phosphate transfection, 20 g DNA was dissolved in 0.5 ml of 0.2 M CaCl 2 then added drop by drop to the 0.5 ml bubbling 2 ϫ HBS (280 mM NaCl, 10 mM KCl, 1.5 mM Na 2 HPO 4 , 12 mM dextrose, 50 mM HEPES). The precipitates were kept at room temperature for 30 min and were then applied to the plates. After 6 h of transfection, the cells were washed with PBS twice and changed to complete medium. Transfections with Lipofectin were done according to the manufacturer's instruction.
Immunocytochemistry-COS-1 cells were grown on cover slips in 24-well plates containing Dulbecoo's modified Eagle's medium with 10% fetal bovine serum and 50 mg penicillin and streptomycin. Before staining, the medium was removed, the cells were washed twice with PBS, fixed with freezing cold 100% methanol in Ϫ20°C for 10 min, and washed again with PBS twice. The cells were then incubated with 5% normal goat serum in PBS for 1 h and incubated in 1:100 dilution of primary antibody in blocking buffer for 1 h. Subsequently, the cells were washed with PBS four times and incubated with 1:100 dilution secondary antibodies labeled with fluorescein isothiocyanate or TRITC (rhodamine) (EY Labs) in blocking buffer for 45 min. Finally, the cells were washed with PBS four times and mounted on the slide with polyvinyl alcohol mounting medium. The slides were air dried and examined with a fluorescence microscope with a 100 watt mercury arc lamp (Nikon, TE300). Images were collected with an interline chargecoupled device camera (Model 1300, Roper Scientific, Trenton, NJ) driven by IP lab software (Scanalytics Inc., Suitland, VA), and assembled in Adobe Photoshop.
The quantification of images was done by assembling fluorescence images from cells transfected with the indicated constructs as well as non-transfected cells (examined by differential-interference contrast microscopy). Images were counted by two individuals blind to experimental condition, and the number and extent of processes for His 6 positive cells (and control cells) was determined.

RESULTS
Construction and Expression of G i ␣ 1 -Transducin Chimeras-To map tubulin-binding sites on G i ␣ 1 we constructed G i ␣ 1 / G t ␣ chimeras where we exchanged several corresponding regions of these two structurally related proteins. The structures of chimeras are schematically illustrated in Fig. 1. The recombinant ␣ subunits were expressed in E. coli and purified using a combination of affinity chromatography on nickel-nitrilotriacetic acid agarose and Mono Q high pressure liquid chromatography. Both the GTP binding capability and the AlF 4 Ϫ induced conformational change as measured by tryptophan florescence at 340 nm were comparable for G i ␣ 1 and each of the 3 chimeric proteins. For G i ␣ 1 , G t ␣, chimera 1, and chimera 2, the AlF 4 Ϫ ⌬F (% increase) was 55, 70, 50 -55, and 60 -70, respectively. For chimera 3, the AlF 4 Ϫ ⌬F (% increase) was 50. The values given represent the increase in fluorescence upon G␣ and G␣ chimera activation. Ranges are given for multiple preparations (between two and six, depending upon the protein). Where a single value is given, two preparations gave the same results. Detected changes in intrinsic Trp fluorescence indicated that G i ␣ 1 and chimeras were fully functionally active. Those values, which are dependent upon the number of tryptophan residues, are similar to the values reported previously for G i ␣ 1 , G t ␣ and G i ␣ 1 /G t ␣ chimeras (21).
Binding of Tubulin by G␣-chimeric Proteins-Previous results had suggested that G i ␣ 1 binds tubulin with a K D of 120 nM, whereas the affinity of G t ␣ for tubulin was too low to be measured reliably (14). A site on G t ␣ thought to interact with effectors was mapped to the residues 237-270 (21). Therefore, two reciprocal chimeras containing the region 237-270 from G i ␣ 1 or G t ␣ (chimeras 1 and 2, respectively) were constructed. Fig. 2 shows that chimera 1 was virtually indistinguishable from G i ␣ 1 in its ability to bind 125 I-tubulin, whereas chimera 2, similar to transducin, did not bind tubulin in any measurable manner. The third chimera, chimera 3, was constructed to measure the contributions of regions 237-270 directly. The replacement of this region of G i ␣ 1 with the analogous region of transducin resulted in a protein that bound tubulin to roughly the same extent as G i ␣ 1 (Fig. 2A). Scatchard analysis of tubulin binding to G i ␣ 1 and chimera 3 reveals saturable binding and a K D of 120 and 123 nM, respectively.
Chimera 3 Is Not Transactivated by AAGTP-Tubulin-The above data show that both chimera 1 and chimera 3 bind to tubulin with about the same affinity as G i ␣ 1 . Chimera 1 can be transactivated via transfer of GTP from tubulin to a similar extent as G i ␣ 1 . Approximately 50% of the AAGTP is distributed to each protein during this process, and GTP in the milieu has no access to either G␣ or tubulin once a complex has formed (19). Fig. 3 demonstrates that chimera 3 failed to serve as a substrate for transactivation despite the fact that it bound to tubulin. When tubulin and G i ␣ 1 were incubated along with chimera 3, transactivation of G i ␣ 1 by tubulin was substantially diminished. Equimolar chimera 3 inhibits transactivation of G i ␣ by 49.9 Ϯ 13%. When the concentration of chimera 3 was increased to twice that of G i ␣ 1 , transfer of AAGTP to G␣ was blocked, and the AAGTP bound to tubulin was also decreased. These data suggest that chimera 3 cannot be transactivated by tubulin, and it inhibits the transfer of GTP from tubulin to G i ␣ 1 . Thus, chimera 3 may act as a dominant negative to inhibit the transactivation of G i ␣ by tubulin.

Chimera 3 Blocks the Formation of Cellular
Outgrowths-Previous studies, as well as data obtained in Figs. 2 and 3, suggested that G␣ might alter cellular projections containing microtubules (11,17,19). To test this, G␣ i1 and chimeras 1-3 were expressed in COS-1 cells. The expression of each of these chimeras, which was about 3-fold greater than the endogenous G i ␣, was identified with an antibody against the His 6 tag on the protein, and the expression level for each one of the constructs appears to be comparable. This was determined by both the immunostaining seen in the Fig. 4 and by Western blot (not shown). Fig. 4 also demonstrates the effect that the expression of the various constructs has on the formation of cellular outgrowths (indicated by arrows).
Native or vector-transfected COS-1 cells extend moderate length cellular processes that tend to be slightly shorter than the body of the cell (Fig. 4 and Table I). Transfection with G i ␣ 1 or chimera 1 increased both the number of cells displaying processes and the mean length of those processes (Table I). Thus, a moderate increase in the expression of G i ␣ increases the length and extent of microtubule-bearing cellular outgrowths. By contrast, chimera 2, which did not bind to tubulin, did not significantly affect cellular outgrowths on COS-1 cells. Both the number and size of processes in chimera 2-transfected cells were similar to control. When cells were transfected with chimera 3, cellular processes were sparse and extremely short ( Fig. 4 and Table I). The expression of chimera 3 prevented the transactivation of endogenous G␣ by tubulin and prevented cells from sending out these microtubule-rich processes (Fig. 4 and Table I). DISCUSSION Data in this study suggest that distinct regions on G␣ mediate the binding and the transactivation of G␣ protein and tubulin. G i ␣ 1 (237-270) plays a crucial role for the transactivation of G i ␣ by tubulin, even though it is not a region required for the binding of those two molecules. Based on the crystal structure of G␣ protein, this region includes the helix of ␣3, a loop, and the ␤5 strand. The (237-270) domain in G i ␣ 1 is one of the binding sites for effectors such as adenylyl cyclase (23). Similarly, this domain in transducin binds the phosphodiesterase ␥ subunit allowing activation of retinal phosphodiesterase (31). The importance of the domain implies that the interaction of G i ␣ 1 and tubulin, which evokes transactivation of the later molecule, may play a role in modulating the physiology of signaling pathway and may orchestrate organization of cytoskeleton.
The ␣ subunits of heterotrimeric G proteins are a family of proteins of 39 -52 kDa that display a similarity of about 45-80% at the amino acid level. Previous studies (14) suggested that G i ␣ 1 bound tubulin with high affinity (K D ϭ 120 nM) and G i ␣ was transactivated by directly transferring GTP. Transducin neither bound tubulin nor was its substrate for transactivation (14). In this study, when the G i ␣ 1 (215-295) segment replaced the G t ␣ sequence, chimera 1 could bind to tubulin and was transactivated by transfer of GTP from tubulin in vitro. This suggests that the region 215-295 might be important for mediating G i ␣ 1 interaction with tubulin. However, in a manner similar to transducin, chimera 2 did not bind tubulin in vitro. Thus, we initially thought that the G i ␣ 1 (237-270) domain might play a key role in rendering G i ␣ 1 able to bind tubulin in vitro. Strangely, chimera 3 bound to tubulin with the same high affinity as wild-type G i ␣ 1 did. Nonetheless, it appears that chimera 3 inhibited the "productive" association between G i ␣ 1 and tubulin, which allows for the transactivation of G i ␣. Previous studies (14,17,24) have shown that there may be multiple binding sites on the tubulin dimer for G i ␣ 1 , and part of the receptor interaction domain of G␣ may mediate this binding. Based on data in this study, it is likely that the G i ␣ 1 (237-270) domain is involved in the binding of G i ␣ and tubulin; however, there is an additional site (or sites) for G i ␣ 1 interaction with tubulin, and this is located within 1-215 and 295-354 regions of G i ␣ 1 . As shown in a solid phase binding assay, replacement of G i ␣ 237-270 with G t ␣ 237-270 in G i ␣ sequence did not significantly affect the binding affinity of chimera 3 and tubulin (K D Х 123 nM for either molecule), implying that this second site(s) is crucial for the high affinity binding of G i ␣ 1 to tubulin.
Tubulin is able to transfer nucleotide (GTP or AAGTP) directly to various G protein ␣-subunits (G s ␣, G i ␣ 1 , and G q ␣) (13,16,19). This phenomenon has been referred to as transactivation, and during the process nucleotide is not released into the milieu but rather is transferred directly from the tubulin to the FIG. 1. Model for G i ␣ 1 /G t ␣ chimera constructs. All constructs contain His 6 tag at the N-terminal. G protein ␣-subunit (19). It suggests that tubulin activates G proteins by directly transferring GTP to G␣, thus "bypassing" a G s ␣-coupled receptor (16). This interaction between G␣ and tubulin has led to a hypothesis that the cytoskeleton may participate in the regulation of G protein-signaling pathways. Colchicine, a microtubule depolymerizing agent, potentiated ␤-adrenoreceptor-stimulated cyclic AMP in lymphoma cells, where activation of the ␤-adrenergic receptor was obligatory for GTP-activation of adenylyl cyclase (25). In permeable C6 cells, tubulin stimulates adenylyl cyclase activity by transactivating G s ␣ (16). Exogenous tubulin can also inhibit rat cerebral cortex membrane adenylyl cyclase, apparently by transactivation of G i ␣ 1 (18,19).
Insertion of G t ␣ 237-270 into G i ␣ did not alter the GTP binding or the characteristics of binding between G i ␣ and tubulin, but chimera 3 did not serve as a substrate for transactivation by tubulin. Chimera 3 also inhibited GTP transactivation of wild-type G i ␣ (Fig. 3). These results suggest that the G i ␣ 237-270 domain is essential to transfer GTP from tubulin to G i ␣. Thus, it appears that chimera 3 acts as a dominant negative mutant of G i ␣ 1 with regard to the "fruitful" interactions between G i ␣ and tubulin. The domain identified might provide a useful tool to further study and understanding the biologic role of interaction of G i ␣ and tubulin in cells.
In COS-1 cells expressing His 6 -G i ␣ 1 and chimeras, it appeared that these proteins were found throughout the cytosol. This may be because the tagged His 6 at the end of the Nterminal of G i ␣ 1 disrupted the G i ␣ 1 associated with plasma membrane (7). Addition of GFP to the N-terminal of G s ␣ blocked its association with plasma membrane (6). However, myristoylated and non-myristoylated G i ␣ bind tubulin with equal affinity and both activate tubulin GTPase, suggesting that they bind to tubulin in a similar fashion (11). Thus, the physiologic potential of these G␣ constructs may be apparent in these experiments. The considerable difference in effect among these varying constructs of G␣ supports this notion. Furthermore, cytosolic G␣ may be more relevant to the modulation of microtubule dynamics than membrane-associated G␣. In Dictyostelium, G␣ appears cytosolic, and this may help to orient cells in response to a polarizing signal (26).
Microtubules are dynamic structures that undergo assembly and disassembly within the cell. They function both to deter-FIG. 2. Specificity and affinity of 125 I-tubulin binding to G␣. A, 125 I-tubulin binding to G protein ␣ subunits. G i ␣ 1 , chimera 1, chimera 2, chimera 3, and G t ␣ were applied to a nitrocellulose sheet in the amounts indicated and air-dried at room temperature. Following this, tubulin binding was assessed by overlay with 125 I-tubulin and autoradiography as described under "Materials and Methods." One of three similar experiments is shown. B, saturation isotherm and Scatchard plot for tubulin binding to G i ␣ 1 and chimera3. Data were derived from dot blotting performed with a method similar to that for A. 100 ng of G i ␣ 1 or chimera 3 were applied to each spot. Triplicate nitrocellulose spots corresponding to the total binding and nonspecific binding (determined in the presence of 100-fold excess unlabeled tubulin) were cut out and counted in an LKB Rack gamma counter. The graph on the left shows the saturation isotherm for specific binding of 125 I-tubulin to chimera 3 (ƒ) or G i ␣ 1 (q). On the right are the Scatchard plots derived form these data. The K D and B max for G i ␣ 1 were 121 nM and 386 fmol/ng, respectively. For chimera 3, these data were 123 nM and 473 fmol/ng, respectively. Data were calculated from two similar experiments.

FIG. 3. Transactivation of G␣ constructs by tubulin.
Freshly prepared tubulin-[ 32 P]AAGTP was incubated with equimolar G i ␣ 1 , chimera 3, G t ␣, G i ␣ 1 plus chimera 3, or twice molar G i ␣ 1 or chimera 3 as indicated. AAGTP binding to tubulin or G␣ was made covalent by UV irradiation, and samples were prepared for SDS-PAGE. [ 32 P]AAGTP bound to tubulin or G␣ (as a result of transactivation) was quantified by phosphoimaging analysis. Images shown are from one of three similar experiments. mine cell shape and in a variety of cell movements, including some forms of cell locomotion, the intracellular transport of organelles, and the separation of chromosomes during mitosis. Several studies (11,19) have suggested G proteins might participate in modulation of the cytoskeleton. Association between tubulin or microtubules and G␣ has been established for some time (14,27,28). In this study, overexpression of His 6 -G i ␣ 1 and chimera 1 in COS-1 cells increases the length and number of cellular processes, whereas chimera 2, which does not bind tubulin, had no effect. This is consistent with the notion that the binding and transactivation of G i ␣ 1 by tubulin might have a role in the regulation of microtubules. Previous studies also show that G␤␥ stabilizes microtubules (4), whereas G i ␣ 1 increases microtubule dynamics by increasing the frequency of microtubule catastrophe (11). Although these data are from in vitro studies, they are consistent with the possibility that G␣ may modulate the dynamics of microtubules in cells. Note that an increase in microtubule dynamics is not inconsistent with increased length and number of cellular processes induced by increased expression of G i ␣ 1 or chimera 1. In fact, microtubules in areas of dynamic cellular extension, such as growth cones, display more dynamic behavior (32). A recent study (33) has demonstrated an increase in the association of G i ␣ (and G s ␣) with microtubules in cellular processes induced by nerve growth factor and other "differentiating" agents in PC12 pheochromacytoma cells.
Curiously, expression of chimera 3 blocks the extension of cellular processes. Although it would be premature to suggest that G i ␣ normally accelerates the extension of cellular process thorough its interaction with tubulin, these results are consistent with such a possibility. To draw more specific conclusions, more studies in vitro and in vivo are required. The role of G␣ transactivation by tubulin in this process is also unclear. Although transactivation seems pertinent to the regulation of adenylyl cyclase or phospholipase C by tubulin (16,20), its role in regulation of microtubule dynamics is unknown. Nevertheless, it is noteworthy that activation of G i ␣ and G s ␣ has been linked to a rapid increase in microtubule depolymerization (29,30).
In summary, this study reveals that distinct regions on G␣ mediate the tubulin binding and the transactivation process. G i ␣ 1 219 -295 is important to the binding of G i ␣ 1 and tubulin, and G i ␣ 237-270 contributes to the transactivation by tubulin of G i ␣ 1 . The data in this study implicate G i ␣ 1 transactivation by tubulin in the regulation of microtubule organization. Such a regulation might provide new insight into the relationship between hormone or neurotransmitter action and cell morphology or other aspects of the dynamic cytoskeleton.

TABLE I
Number and extent of cell process formation in COS 1 cells expressing various chimeric G protein constructs COS-1 cells were transfected (calcium phosphate) with constructs indicated and then were fixed with Ϫ20°C methanol and stained with rabbit anti-His 6 tag and mouse anti-␣-tubulin (ICN). Forty to forty-five cells were counted for each construct from 3 different transfections. A process was defined as an extension from the cell body longer than one third of the cell length. Values are mean Ϯ SEM for 40 -45 cells from three separate transfection experiments. Both the number of processes/ cell and the average process length are significantly (p Ͻ 0.001) less for cells transfected with chimera 3 than observed in control cell.  4. G protein and tubulin dual staining of transiently transfected COS-1 cells. COS-1 cells were transfected with 10 g of pcDNA3 coding for G i ␣ 1 , chimera 1, chimera 2, and chimera 3, respectively, and grown in normal Dulbecoo's modified Eagle's medium for 48 h. The cells were fixed with Ϫ20°C methanol and double stained with polyclonal His 6 tag antibody and fluorescein isothiocyanate goat anti rabbit IgG and with ␣-tubulin antibody and rhodamine-coupled goat anti-mouse IgM. G␣ constructs appear green and microtubules are seen in red. The figure shows the overlay of the tubulin and His 6 tag antibodies. Areas of tubulin/G␣ overlap appear yellow. The arrows indicate cellular processes. Images shown are typical of five experiments in which at least 60 cells of each type were examined.